Abstract
The pharmacokinetics (PK) of biologic therapeutics, especially monoclonal antibodies (mAbs), in monkeys generally presents the most relevant predictive PK information for humans. However, human mAbs, xenogeneic proteins to monkeys, are likely to be immunogenic. Monkeys previously treated with a human mAb (non-naïve) may have developed antidrug antibodies (ADAs) that cross-react with another test mAb in subsequent studies. Unlike PK studies for small-molecule therapeutics, in which animals may be reused, naïve monkeys have been used almost exclusively for preclinical PK studies of biologic therapeutics to avoid potential pre-existing immunologic cross-reactivity issues. The propensity and extent of pre-existing ADAs have not been systematically investigated to date. In this study, the PK and immunogenicity of mAb A, a human anti-human interkeukin-17 mAb, were investigated in a colony of 31 cynomolgus monkeys previously exposed to other human mAbs against different targets. We screened the monkeys for pre-existing antibodies to mAb A prior to the PK study and showed that 44% of the monkeys had pre-existing cross-reactive antibodies to mAb A, which could affect the PK characterization of the antibody. In the subcolony of monkeys without measurable pre-existing ADAs, PK and immunogenicity of mAb A were successfully characterized. The impact of ADAs on mAb A PK was also demonstrated in the monkeys with pre-existing ADAs. Here we report the results and propose a pragmatic approach for the use of non-naïve monkeys when conducting PK studies of biologic therapeutics.
Introduction
A thorough understanding of the pharmacokinetics (PK) in preclinical species provides critical information for the prediction of PK in humans, as well as for the projection of a safe and sensible starting dose prior to first-in-human clinical trials. Results from numerous studies (e.g., Wang and Prueksaritanont, 2010; Dong et al., 2011; and Oitate et al., 2011 and 2012) suggested that the PK in monkeys presents the most relevant information for the prediction of PK of biologic therapeutics, especially monoclonal antibodies (mAbs), in humans.
The Fc domain of IgG binds to the neonatal Fc receptor (FcRn), which protects the IgG from endosomal proteolysis following endocytosis in the protein catabolism process (Kuo and Aveson, 2011), except for IgG3, which has substantially lower FcRn binding compared with other IgGs (Stapleton et al., 2011). The IgG-FcRn binding affinity and specificity of human mAbs in nonhuman primates are very similar to those in humans. PK results from monkeys can very reasonably predict PK in humans (i.e., predicted clearance within 2-fold of the clinical observation), primarily resulting from the similarities in the FcRn salvage mechanisms between monkeys and humans (Han and Zhou, 2011). However, human mAbs, acting as xenogeneic proteins when administered to monkeys, are likely to be immunogenic.
The impact of antidrug antibodies (ADAs) on PK and pharmacodynamics (PD) has been well documented (Richter et al., 1999; Lobo et al., 2004; Wang et al., 2008; Vugmeyster et al., 2010), with the primary result being enhanced clearance of the immune complex. Furthermore, monkeys that have been treated with a human mAb may have developed cross-reactive ADAs that could bind other test human mAbs administered in subsequent studies. The impact of pre-existing ADAs on PK could be significant, but is also hard to predict because the incidence of ADAs is variable between individual studies and among biologic therapeutics. Unlike PK and PD studies of small-molecule therapeutics, in which animals are often reused in subsequent studies for other test molecules, studies of biologic therapeutics require the use of protein therapeutics–naïve animals to minimize potential immunologic cross-reactivity from pre-existing ADAs and obtain reliable results (Vugmeyster et al., 2008, 2009; Yeung et al., 2010). With administration of xenogeneic proteins, animals that develop an ADA response typically maintain memory cells that result in a more florid ADA response with the introduction of the antigen a second time (Goronzy and Weyand, 2012). This situation is not amenable to washing out of the therapeutic. Cross-reactive ADAs would follow this path as well. Naïve monkeys have been used almost exclusively for preclinical PK studies of biologic therapeutics; however, the issues concerning the occurrence and impact of pre-existing and potential cross-reactive ADAs have not been investigated and documented thoroughly. Currently, >650 biologic medicines, not including vaccines, are in development and >50% of the biologics are mAbs (Pharmaceutical Research and Manufacturers of America, 2013). Thirty of 44 (68%) therapeutic mAbs approved or in review in the European Union and United States are of the IgG1 isotype (http://www.antibodysociety.org/news/approved_mabs.php). As a result of this, the demand for naïve monkeys has been increasing and continues to increase with the growth in the number of biologic therapeutics in drug development worldwide. This demand has created a challenge for researchers, who have to ensure the generation of reliable preclinical data while upholding the “3R” (replacement, reduction, and refinement) ethical principles pertaining to the use of laboratory animals.
With this challenge in mind, a study was designed to quantify the risks and the impact of pre-existing cross-reactive ADAs in a human mAb PK study using a non-naïve monkey colony, and to explore a feasible approach to employ such colonies for PK or PD studies of biologics. In this study, the PK and immunogenicity of mAb A, a human anti-human interleukin (IL)-17 mAb, were investigated in a non-naïve cynomolgus monkey colony that had been previously treated with other human mAbs against different targets. The monkeys in the colony were screened for pre-existing cross-reactive antibodies to mAb A prior to the PK study, and a subcolony of monkeys without measurable pre-existing ADAs was selected for inclusion in the study. The PK and immunogenicity of mAb A were successfully studied in the subcolony of monkeys that did not have pre-existing ADAs, and the results are reported in this paper. Additionally, screening results showed a significant incidence of pre-existing ADAs in the non-naïve monkey colony. The adverse impact of pre-existing cross-reactive ADAs to mAb A on the PK of mAb A was further evaluated and quantified. The results from this study demonstrated a successful approach to conducting a human mAb PK study in non-naïve monkeys that had been treated previously with similar biologic therapeutics. The methodology used in this study demonstrates a strategy and pragmatic approach for conducting PK studies using non-naïve monkeys, addressing the concern of dealing with cross-reactive ADAs while maintaining the principles of 3R in the use of laboratory animals.
Material and Methods
Test System.
Thirty-two non-naïve, healthy, male cynomolgus monkeys (Macaca fascicularis) provided by Hainan Jingang Laboratory Animal Co. Ltd. (Hainan, China) were selected for the study. All animals had been used once in a previous single-dose PK/PD study in which different human IgG1 antibodies were administered. The in-life experimental portion of the current study was conducted at WuXi AppTec (Suzhou, China). The monkeys ranged in age from 3–5 years old and weighed 3–5 kg. The animals were acclimated to the study environment for at least 2 weeks prior to dose administration. During the acclimation and study period, the animals were individually housed according to the standards set forth by the Animal Welfare Act (United States) and in compliance with the Guide for the Care and Use of Laboratory Animals. The animal rooms were controlled and monitored for temperature (19–26°C), humidity (40%–70%), and a 12-hour light/dark cycle, except for necessary brief interruptions for study activities. The animals were fed ad libitum with certified primate diet (Beijing Keao Xieli Feed Co., Ltd., Beijing, China). Fruit was provided daily for nutritional enrichment, and occasional treats were allowed. Purified drinking water was provided ad libitum. The study protocol, including animal husbandry, study design, and in-life procedures, was reviewed and approved by the Institutional Animal Care and Use Committee at WuXi AppTec.
Test Article.
mAb A is an IgG1-based human mAb that specifically neutralizes human IL-17 and binds with similar affinity to monkey IL-17. mAb A was formulated in solution containing 13 mM histidine, 8.5% sucrose, and 0.04% polysorbate 80 at pH 5.4. The concentration of the original formulation was 50 mg/ml. Dosing solutions of mAb A were prepared just prior to administration by serially diluting the original formulation using 5% dextrose solution for injection to yield the desired final concentrations. The dose, concentration, and dosing volume for each treatment group are summarized in Table 1.
Dose and dosing solution for each treatment group
Study Design.
The study was designed as a two-part study. The animals were initially screened for the presence of potential pre-existing cross-reactive ADAs in part 1, and the PK characterization of the test article (mAb A) was conducted in part 2 of the study.
In part 1 of the study, ∼2 months prior to starting the PK study (part 2), one blood sample from each prospective study animal was collected for ADA testing. Serum samples were assayed for ADAs to mAb A. The test samples were collected ∼2 months following the completion of a previous PK study and ∼3 months after the previous mAb dose administration.
In part 2 of the study, the monkeys were grouped based on the ADA test results obtained in part 1. Monkeys that tested negative for pre-existing cross-reactive ADAs to mAb A were grouped into a negative subcolony for part 2 of the study, whereas monkeys that were positive for pre-existing ADAs were considered to be at risk of adversely affecting the PK study and were grouped into an ADA-positive subcolony. The animals from the two subcolonies were separately and randomly assigned into the four treatment groups indicated in Table 2, such that there were at least four monkeys without pre-existing cross-reactive ADAs in each of the dosing groups. Animals in groups 1, 2, and 3 received a single intravenous injection of 1, 3, or 10 mg/kg mAb A, respectively. The animals in group 4 received a single subcutaneous injection of 3 mg/kg mAb A. The day of dosing (beginning of the PK study) was designated as study day 1.
Animal assignment based on pre-existing ADA test results at time of screening
One ADA-negative animal was not assigned due to skin conditions.
Sample Collection and Handling.
At predetermined time points, ∼1.5 ml (or 2.5 ml when the sample was collected for both PK and ADA assessment) of whole blood was collected from each animal into a prelabeled Vacutainer (Becton, Dickinson and Company, Shanghai, China) containing no anticoagulant. The blood samples were allowed to coagulate at room temperature for at least 30 minutes but no more than 1 hour. The samples were then centrifuged at 2800g for ∼15 minutes at ambient temperature to prepare the sera. The serum samples were then frozen on dry ice and stored at −70°C or below in individually labeled tubes until bioanalysis was performed.
One sample from each monkey collected in part 1 was assayed for pre-existing ADAs to mAb A. In part 2, samples were collected on days 1 (predose and 0.5, 1, 2, and 6 hours postdose), 2, 3, 4, 8, 15, 22, 29, 36, 43, and 51 from the animals dosed via intravenous administration. For the animals dosed via subcutaneous administration, samples were collected on a similar schedule except that the 0.5-, 1-, and 2-hour postdose samples on day 1 were omitted. The serum samples were assayed for mAb A concentration. In addition, day 1 predose and day 51 samples were assayed for ADAs to mAb A and the concentration of free IL-17.
Determination of Serum mAb A Concentration.
A validated specific electrochemiluminescence (ECL) immunoassay was used for the determination of serum concentration of mAb A. Capture and detection reagents are developed from two different anti-idiotypic antibodies. In brief, the assay was conducted on streptavidin-coated 96-well plates. The assay plates were blocked for 1 hour with buffer. After washing, the test samples, standard curve calibrators, and quality control samples were added to the plates along with biotinylated capture antibody. The plates were then incubated for 1 hour at room temperature. Ruthenium-labeled detection antibody was then added to each plate well. Following a 1-hour incubation, the plates were washed, substrate was added, and the resulting ECL signal was read on a Sector Imager 6000 reader (Meso Scale Discovery, Gaithersburg, MD). The concentration of mAb A was determined by interpolation from the mAb A calibration curve. The lowest quantifiable concentration in a sample was 0.08 μg/ml.
Determination of ADAs to mAb A.
Validated ECL immunoassays were used for the bioanalysis of monkey serum samples. The analyses of ADAs to mAb A comprised a series of three types of ECL bridging immunoassays: screening assay, specificity assay, and titration assay. Briefly, streptavidin-coated plates equipped with well-bottom electrodes from Meso Scale Discovery were used. For initial sample screening (screening assay), study samples incubated with ruthenium-labeled mAb A and biotinylated mAb A were captured on the plate. An ECL cut point was set based on the 95th percentile of baseline data obtained in 50 naïve healthy monkey serum samples. A sample was considered to be potentially positive when the ECL was ≥114. Samples classified as potentially positive were subsequently tested in a confirmatory test (specificity assay). To determine if a potential ADA-positive sample was specific to mAb A, percent inhibition of ECL in the presence of 200 μg/ml mAb A was determined. A cut point of 23.9% was preset based on the 99.9th percentile of the percent inhibitions obtained in 25 naïve monkey serum samples. The confirmatory test also examined potential assay interference by IL-17 in the presence of goat anti-human IL-17 antibody (cross-reactive with monkey IL-17). Positive sample was confirmed if the following conditions were met: 1) in the presence of mAb A, the percent inhibition was ≥23.9% or the assay result was <114 ECL; and 2) the assay result was ≥114 in the presence of goat anti-human IL-17 antibody. A sample was classified as negative if the assay result was <114 ECL in the presence of the goat antibody. Positive samples were also serially diluted (titration assay) to obtain a relative measurement of serum antibody concentration against mAb A. For the titration assay, an ECL value of ≥250 ECL was considered as positive for antibodies to mAb A. Titer was determined only for the positive samples collected in part 2 of the study. A mouse monoclonal anti–mAb A antibody was used as a positive control for the consistency and specificity tests. The sensitivity of the assay at the cut point was evaluated using cynomolgus monkey polyclonal ADAs at 195 ng/ml, or at 49 ng/ml using positive control mouse mAb. The detection of low positive ADA control could tolerate up to 3 μg/ml mAb A in a sample. It should be pointed out that residual drug concentration that exceeds the tolerance limit of the ADA assay may interfere with the detection of ADAs and potentially result in a false-negative assay outcome for the sample.
Determination of Serum Free IL-17 Concentration.
Serum concentration of IL-17 was analyzed using a modified enzyme-linked immunosorbent assay from a commercially available kit (MabTech Inc., Cincinnati, OH). Free monkey IL-17 concentration was determined using the capturing antibody and biotinylated detection antibody provided in the kit. In brief, assay plates were incubated overnight with capture antibody in phosphate-buffered saline. The plates were washed and blocked for 1 hour using 0.1% bovine serum albumin in assay buffer. Diluted test serum samples or standards were added, and the plates were incubated for 2 hours. Detection antibody was then added and followed by a 1-hour incubation. After a wash step, diluted streptavidin–horseradish peroxidase was added to each well and the plates were incubated for 1 hour, followed by addition of tetramethylbenzidine substrate. All incubations were done at room temperature. The reaction was stopped by the addition of 2 N H2SO4. Spectrometric absorbance was read at 450 nm with 570-nm subtraction. In the presence of IL-17–mAb A complex, the assay was not capable to accurately quantify free IL-17 concentration.
Pharmacokinetic Analysis.
The serum concentration–time profile of mAb A for each individual animal was evaluated by noncompartmental analysis using WinNonlin (V5.1; Pharsight, Mountain View, CA). The Cmax, area under the serum concentration–time curve (AUC), volume of distribution (Vz), clearance (CL), and half-life (t1/2) were calculated when data allowed. The pharmacokinetic analysis and descriptive statistics were performed separately for the monkeys from the positive and negative subcolonies because the terminal phase in the ADA-positive animals may be poorly defined and the related PK parameters may not be accurately estimated.
Results
Screen Results for Pre-existing Cross-Reactive ADAs to mAb A
One serum sample from each of the 32 prospective animals was collected for the assessment of pre-existing cross-reactive ADAs to mAb A in part 1 of the study. Fourteen of the 32 samples tested positive for pre-existing antibodies to mAb A. The results of two positive samples are worth mentioning. The sample from animal #707625 had an ELC value of 114 and met the screening assay cut point of ≥114. The sample from animal #706559 met the screening assay cut point, but barely missed the specificity assay cut point of ≥23.9% inhibition with a 23.2% inhibition. To make sure a relatively “clean” subcolony was selected for part 2 of the study, these two animals were classified as ADA-positive for part 2 of the study. As a result, the positive rate for pre-existing antibodies to mAb A was 44% (14 of 32 animals) in the non-naïve monkey colony. Of the 18 animals that tested negative for pre-existing ADAs, one was excluded from the study due to a skin condition that developed before the initiation of part 2. Therefore, the non-naïve monkey colony was separated into two subcolonies, with 17 animals in a subcolony without pre-existing ADA and 14 in a subcolony with pre-existing ADA. As shown in Table 2, the monkeys from the two subcolonies were separately and randomly assigned to the four dosing groups according to the part 2 study design. The ADA status of individual animals is summarized in Table 3.
Summary of ADA test results at different stages of the study with historical exposure information
The 31 study animals had previously received a single administration of one of the following human mAbs: mAb 1, mAb 2, mAb 3 (a humanized anti–respiratory syncytial virus IgG1 κ monoclonal antibody), mAb 4 (mAb 3 with an Fc allotypic variant), mAb 5 (mAb 3 with a mAb 1–like light-chain construct), mAb 6 (mAb 3 with a mAb 2–like light-chain construct), mAb 7 (similar to mAb 1 with a modified light chain), or mAb 8 (similar to mAb 2 with a modified light chain). Among the previously administered mAbs, mAb 1 was a human IgG1 anti-human IL-12/23 p40 antibody and mAb 2 was a human IgG1 antibody neutralizing human tumor necrosis factor-α. The other six were human IgG1κ mAbs; however, there were Vκ variants, different heavy chain/light chain combinations, and allotypic variants in the panel. mAb 3 had additional modifications to the heavy chain. mAb 7 and mAb 1 had one mutation in the heavy chain that differed from the other mAbs. Thus, mAb A has a heavy-chain constant region sequence identical to five of the eight mAbs, namely, mAbs 2, 4, 5, 6, and 8. The unique variable domain for mAb A is the λ light chain. mAb 3 and mAb 4 do not bind to any known monkey target. mAbs 5 through 8 theoretically bind to the same target as their respective parent molecules but were not fully tested. The t1/2s of these mAbs were from 3.4 days to 13.5 days estimated from the previous study. The prior mAb exposure history of each animal used in the study is also indicated in Table 3. There was no apparent correlation between the presence of pre-existing ADAs and the type of mAb the animal received in the previous study.
The results from the specificity assay are presented in Table 5. In the presence of excess mAb A, the percent inhibition of ECL result ranged from 23.2%–99.8%. To characterize the immunogenic cross-reactivity, an effort was made to conduct the specificity assay using the original antigens—previously administered mAbs. The percent inhibition was determined in the presence of the previously administered mAbs instead of using mAb A. A cut point, however, was not predetermined for the test. Additionally, being limited by the availability of the mAbs used in the previous study and residual sample volume for screening, the test was run only with mAb 1 and mAb 2 for all samples, and mAb 7 and mAb 8 only for the samples from animals dosed with respective antibodies. The results are summarized in Table 4. In general, the percent inhibition in the presence of originally administered mAbs correlated very well with that in the presence of mAb A based on the limited data. The results suggested that the ADA was cross-reactive.
Summary of specificity assay results in part 1
Summary of specificity and titration assay results for all ADA-positive monkeys in part 2
Pre-existing ADAs to mAb A and Treatment-Related Immunogenicity of mAb A
Animal ADA Status.
Samples were collected on day 1 (predose) and day 51 from the 31 animals randomized into part 2 of the study. These samples were analyzed for antibodies to mAb A after completion of the in-life portion of the PK study. The monkeys showed some minor variability in ADA status during the period between part 1 and the initiation of part 2. Prior to mAb A administration, 14 of the 17 animals (82%) from the negative ADA subcolony remained negative. Three animals (#705599, #707593, and #706571) in the negative ADA subcolony became positive for antibodies to mAb A during the period of ∼2 months between parts 1 and 2.
In the positive ADA subcolony, the results from the samples collected on day 1 (predose) showed that 11 of 14 animals (79%) remained positive prior to the administration of mAb A. The two animals (#707625 and #706559) from the positive ADA subcolony that had positive test results near the assay method sensitivity limit became negative. In addition, animal #705539 had a positive screen result (ECL = 118) but did not meet the specificity cut point with a 21% inhibition. ADA status results in the predose samples are summarized in Table 3.
Following mAb A treatment of the monkeys in the negative ADA subcolony, 15 of 17 (88%) tested negative for antibodies to mAb A. Of the three formerly ADA-negative animals that became positive for ADAs prior to the administration of mAb A, animal #706571 tested negative following mAb A treatment, while the other two animals remained positive. mAb A was not immunogenic in the 14 animals that tested negative for pre-existing ADAs in part 1 and remained negative for ADAs through part 2 (Table 2) of the study. mAb A was also considered to be nonimmunogenic in one additional monkey (#706571) that tested positive for ADAs prior to mAb A administration, but negative at the end of the study. However, the immunogenicity of mAb A could not be determined in the two animals that tested positive for ADAs at the end of the study because their predose samples also tested positive for ADAs.
Following mAb A treatment in monkeys from the positive ADA subcolony, 8 of 14 (67%) were positive for ADAs to mAb A. There were 11 animals positive for ADAs prior to the administration of mAb A. Two animals (#707625 and #706559) in the positive ADA subcolony that had become negative for ADAs prior to mAb A administration, mentioned earlier, remained negative at the end of the study. One animal (#705539) that did not meet the specificity cut point was also negative at the end of the study. Three additional ADA-positive animals (#705609, #706545, and #706575) became ADA-negative at the end of the study. The initial ADA response in these animals was not treatment-responsive, and therefore, mAb A was not considered to be immunogenic in these six animals. The immunogenicity of mAb A could not be determined in the remaining eight ADA-positive animals because both their screening and pre-dose samples tested positive for ADAs.
ADA-Positive Sample Confirmation and ADA Titer.
Serum samples from the animals in the positive ADA subcolony and from the two monkeys in the negative ADA subcolony that tested positive at the end of the study were investigated in sequential confirmatory assays. Preincubation with excess goat anti-human IL-17 antibody did not significantly alter the signal or the conclusions. Preincubation with an excess concentration of mAb A inhibited the assay signals to different extents in different samples. The results from the specificity assay are summarized in Table 5.
In part 2 of the study and prior to mAb A treatment, the percent inhibition ranged from 37.3%–95.4%. In the samples collected on day 51 following administration of a single dose of mAb A, the percent inhibition ranged from 69.4%–97.3%.
To obtain a quasi-quantitative measurement of ADA level, the titer of antibodies to mAb A in all ADA-positive samples collected in part 2 of the study on day 1 (predose) and day 51 were evaluated in the titration assay. The results are summarized in Table 5. ADA titer prior to the administration of mAb A ranged from 20–5120 (Table 5, Predose). The titer was not determined in seven samples for which ECL values were <250 (the titration assay method cut point) at the minimal required dilution of the assay. After completion of the in-life portion of the study, ADA titer ranged from 20–40,960 (Table 5, End of Study), with a median of 1920. In all but two animals (#707593 from the negative ADA subcolony and #706545 from the positive ADA subcolony that had an ADA-negative result at the end of the study), the titer generally increased following administration of mAb A. This indicates that most animals with pre-existing ADAs to mAb A showed treatment-responsive increases in ADA titer after administration of the drug and only a minority of the animals with pre-existing ADAs were unresponsive to the administration of mAb A.
The PK of mAb A in Non-Naïve Monkeys
The PK of mAb A was evaluated separately for the ADA-negative and ADA-positive subcolonies. The mean concentration-time profiles of mAb A in animals from the negative ADA subcolony are presented in Fig. 1. The profiles of each animal were analyzed to estimate the PK parameters of mAb A in cynomolgus monkeys. Following a single intravenous administration of mAb A, the Cmax and AUC from time 0 to infinity (AUCinf) increased in an approximately dose-proportional manner. The average CL of mAb A ranged from 3.99–5.16 ml/day per kg for the different dose groups. The volume of distribution of the drug appeared to be shallow, similar to that of other therapeutic mAbs or proteins with high molecular weight, and was estimated at 86.0–91.3 ml/kg for the different dose groups. The elimination t1/2 ranged from 12.4–15.6 days. A summary of the mean PK parameters following intravenous administration is presented in Table 6. The CL, Vz, and t1/2 values are relatively consistent across the different dose groups. These PK parameters are consistent with those obtained from a different study using naïve monkeys, also shown in Table 6.
Mean serum mAb A concentration-time profiles following a single-dose administration of 1, 3, and 10 mg/kg i.v. (A) or 3 mg/kg s.c. (B). The profiles of individual animals that tested positive for ADAs to mAb A (indicated by the animal’s ID) are also plotted, along with the mean profile of the dose group for comparison. The lowest quantifiable concentration of mAb A in a sample was 0.08 μg/ml.
PK parameter estimates (mean ± S.D.) following a single intravenous dose of mAb A in cynomolgus monkeys from the subcolony without pre-existing ADA
Following a single subcutaneous administration of 3 mg/kg mAb A, an average Cmax of 23.7 μg/ml was attained in about 3 days post–dose administration (Tmax in Table 7). The bioavailability of mAb A was estimated to be ∼93% based on a comparison of the AUCinf to that following intravenous administration of 3 mg/kg mAb A.
PK parameter estimates (mean ± S.D.) following a single subcutaneous dose of mAb A in cynomolgus monkeys from the subcolony without pre-existing ADA
Except for the determination of Cmax, the two animals (#705599 and #707593) that tested positive for ADAs to mAb A were excluded from the calculation of the mean (S.D.) PK parameters. The individual PK parameters for the two animals are listed separately in Table 8. The concentration-time profiles of the two animals are plotted in Fig. 1 for a comparison with the mean concentration profile. Although animal #706571 was ADA-positive prior to the dose administration, it was considered to be ADA-negative for the study because the sample from day 51 showed a negative ADA result. The PK results from this animal were included in the descriptive statistical summary.
PK parameter estimates following a single dose of mAb A in ADA-positive cynomolgus monkeys from the subcolony without pre-existing ADA
The individual concentration-time profiles of mAb A in animals from the subcolony with pre-existing cross reactive ADA are presented in Fig. 2. PK parameters determined from the concentration-time profiles are summarized in Table 9. Following a single intravenous administration, the mean Cmax increased in an approximately dose-proportional manner. However, an accelerated decrease in mAb A concentration during the elimination phase was observed in all eight ADA-positive animals. A poorly defined elimination phase prevented an accurate estimation of PK parameters in these animals. The remaining six animals from the positive ADA subcolony that were ADA-negative at the end of the study had very similar PK parameter values to those obtained in animals from the negative ADA subcolony.
Individual serum mAb A concentration-time profiles following a single-dose administration of 1 mg/kg i.v. (A), 3 mg/kg i.v. (B), 10 mg/kg i.v. (C), or 3 mg/kg s.c. (D) in cynomolgus monkeys from the subcolony with pre-existing ADA. Each symbol-line represents one individual animal in the dose group. The mark (#) next to an animal’s ID number indicates that the animal was negative for ADAs to mAb A at the end of the study.
PK parameter estimates following a single dose of mAb A in cynomolgus monkeys from the subcolony with pre-existing ADA
Relationship between ADA and Impact on PK
Only one (animal #705599) of the two animals with a positive ADA result in the negative subcolony showed an appreciably faster decrease in mAb A concentration during the elimination phase. The t1/2 for this animal was 5.71 days, much shorter than the mean t1/2 of 12.4–15.6 days obtained from the ADA-negative subcolony. The value of AUCinf in this animal was lower, but to a less significant extent. The PK parameters for the other ADA-positive animal appeared to be consistent with those from the ADA-negative animals. Relatively low titer was determined in both ADA-positive animals from the negative ADA subcolony. The individual PK parameters for the two ADA-positive animals are listed separately in Table 8.
In animals from the positive ADA subcolony, the PK was markedly affected in the eight monkeys that tested positive for ADAs at the end of the study. These animals had relatively high titers of ADAs to mAb A. The accelerated decrease in concentration over time brought the profile down to lower than the lowest quantifiable concentration much earlier than expected. The impact on the PK in ADA-positive animals was much more pronounced in the subcolony with pre-existing ADA.
Baseline Circulating IL-17 Levels
Circulating free IL-17 concentration was measured in the day 1 (predose) samples. In general, baseline concentrations of circulating IL-17 were variable and low. The concentration ranged from undetectable to 20 pg/ml. There was no difference in average IL-17 concentration between the ADA-negative and -positive subcolonies based on IL-17 levels of 2.97 ± 4.93 versus 2.93 ± 3.08 pg/ml for the ADA-negative and -positive subcolonies assigned by part 1 screening results, respectively, and levels of 3.18 ± 5.01 versus 2.69 ± 2.91 pg/ml for the predose ADA-negative and -positive animals determined in part 2 of the study, respectively. Samples collected on day 51 were also tested for free IL-17 concentration; however, the assay was not capable of differentiating free IL-17 from mAb A–IL-17 immune complex. As a result, free IL-17 concentration in the day 51 samples is not reported.
Discussion
Human mAbs, acting as xenogeneic proteins, are likely to be immunogenic in monkeys. The incidence and impact of ADAs on PK may vary among individual animals and different mAbs. The incidence of ADAs could range from a few animals to almost 100%, based on personal observation. Although thought to be likely, it was unclear whether previous exposure to a therapeutic protein would lead to potential immunologic cross-reactivity with a similar, but different, protein in a subsequent PK study. Due to the similarities among some classes of therapeutic proteins (e.g., mAbs), there have been general concerns of immunogenic cross-reactivity of previously developed ADAs; thus, protein-naïve monkeys were used for PK studies of mAbs and other therapeutic proteins (Vugmeyster et al., 2008, 2009; Yeung et al., 2010). Using naïve nonhuman primates presents an increased demand for animal resources, which are often intensively reviewed under the scope of the ethical use of laboratory animals and animal welfare in addition to increasing the cost of drug discovery and development. Nonetheless, few publications have provided data from systematic investigations in this regard. That would allow for the reduction or minimization of the use of animals for these types of studies.
In this study, the PK and ADA for a human mAb were investigated in non-naïve cynomolgus monkeys. The results showed that a large percentage (44%) of non-naïve monkeys had pre-existing cross-reactive antibodies to mAb A. Although the serum concentrations of the previously administered mAbs were not determined as part of the current study, the concentrations were likely below the lowest quantifiable limit during this study based on data from the prior studies and the time elapsed (i.e., >8 half-lives). Despite the small sample sizes associated with each of the previously administered mAbs, there was no apparent correlation between ADA occurrence and the type of mAb previously administered. Because the Fc regions were very similar and the Fab domains were variable among the eight previously administered mAbs and mAb A, a hypothetical epitope(s) could be present on the heavy chain or other shared human sequences and serve as a common site(s) for monkey ADA recognition. The results generally confirmed the concern of immunogenic cross-reactivity, knowing that the extents may vary by the structure, functions, and similarity of historical and subsequent test molecules.
The detection of ADAs and evaluation of the impact on PK are complex (Geng et al., 2008). It was previously reported that human IL-17A/F may form hetero- or homodimers (Wright et al., 2007) at higher concentrations (i.e., nanograms per milliliter). Theoretically, the IL-17 dimer may interact with ADA assay reagents and generate a false-positive signal. To address this potential concern, basal serum IL-17 concentration was measured in the study animals. The concentrations were variable among the individual animals, but were in the picograms per milliliter range. No apparent correlation between IL-17 concentration and ADA positivity was observed. An additional step using goat anti-IL-17 antibody was also incorporated into the ADA assays used in this study to ensure accurate detection.
Pre-existing ADAs were confirmed in the specificity assay in which the ECL signal was inhibited in the presence of excess mAb A. The animals were previously exposed to one of eight different human IgG1 mAbs that were very similar to mAb A. Limited data suggested that the pre-existing ADAs were cross-reactive. The pre-existing ADAs are likely monkey polyclonal antibodies to human IgG and, therefore, cross-reactive to human IgG molecules, rather than specific only to mAb A. Increased ADA titer following mAb A treatment suggested that the treatment boosted pre-existing ADAs and/or was more immunogenic in the animals with the pre-existing ADAs. Increasing titer following treatment is a hallmark of treatment-responsive pre-existing ADA responses (Shankar et al., 2014).
When exploring the potential use of non-naïve monkeys in a study, two basic approaches can be considered: dose more animals than needed and only use the data from ADA-negative animals based on end-of-study results; or prescreen the prospective study animals to select those that are negative for pre-existing cross-reactive ADAs. The pros and cons associated with the resources, drug substance requirements, timing of operations, and potential study risks are obvious for each approach. The results from this study show that both approaches are viable. We think that screening prospective animals prior to the initiation of a study is more practical and carries relatively less risk. The incidence of positive pre-existing ADAs in this study should give an approximate estimate of the initial number of prospective non-naïve animals needed for a study. The incidence of ADAs may vary depending on the exposure history of the animal and the test article administered. No data are available to date for animals treated more than once with similar test articles in sequential studies. The screening approach in this study is effective. Some minor variations were expected because only one prestudy screening sample was collected from each animal. Knowing that there is a risk that a small number of “false-negative” animals may be included in a preclinical PK study, proper sampling for ADA evaluation and using a reasonable group size should be considered as a mitigation plan.
The timing of the prestudy ADA testing sample collection was chosen to be after a representative washout period of ∼2 months, when animals could typically be reused. A longer washout may be necessary to clear residual drug for those mAbs with extended half-lives (Oganesyan et al., 2009; Zalevsky et al., 2010). As expected, there were some minor variations of animal ADA status observed during the ∼2-month time interval between part 1 screening and the initiation of part 2. These variations may slightly affect the selection of animals for inclusion in the study. In addition, the ADA results from part 2 predose samples suggested that a longer washout period, at least in the range investigated, would not provide any extra benefit for clearing pre-existing immunologic cross-reactivity in non-naïve monkeys.
The PK of mAb A was successfully characterized in selected non-naïve monkeys. The PK properties of mAb A and its low immunogenicity in monkeys were consistent with the results obtained from other studies in which naïve monkeys were used. In addition, when ADAs developed in the study animals, the impact on PK in individual monkeys was similar to that seen in other studies.
Conversely, the PK results were affected significantly (Fig. 2) in all eight ADA-positive monkeys from the subcolony with pre-existing ADA. In these animals, serum concentrations of mAb A decreased abruptly, which hindered defining the terminal phase of the concentration-time profile. A faster clearance associated with the formation of a drug-ADA immune complex was suggested as a possible mechanism (Rojas et al., 2005). Although the clearance process has not been fully understood, more efficient phagocytosis due to the larger molecular size of the complex or Fcγ receptor–mediated clearance has been postulated (Uchida et al., 2004). Whether this finding was attributable to treatment-induced or treatment-boosted ADAs in this study is unclear. Increased ADA titers following the mAb A administration strongly suggested that these were treatment-responsive pre-existing ADA responses. Intriguingly, when the end-of-study samples tested negative for antibodies to mAb A, the PK results appeared to be unaffected in these animals even though they were associated with the subcolony with pre-existing ADA.
Despite a limited number of animals and data from each treatment group, there was no correlation between dose or route of administration and ADA status or titer observed in this study.
It has been mentioned earlier that mAb A and the eight previously administered mAbs are all IgG1-based. One should possibly expect different results when another type of IgG-based therapeutics, such as IgG2 and IgG4, or different protein platform is involved. Until more data on a wider variety of molecules are available, this study can be considered as a case study. However, the pragmatic approach and underlying thoughts exemplified in this study should be helpful for further exploration of reduction of the use of naïve animals.
In conclusion, the risk of immunologic cross-reactivity from pre-existing ADAs in non-naïve monkeys used in a sequential PK study was confirmed and quantified. Selection of a suitable animal study pool based on an assessment of pre-existing ADA status prior to initiation of a PK study has proven to be efficient and accurate. Consequently, the PK properties and immunogenicity of mAb A, a human anti-IL-17 mAb, were successfully characterized in a selected subcolony of animals that tested negative for pre-existing ADAs to the test mAb. In addition, this case study provides a feasible and practical approach for using non-naïve monkeys in preclinical PK and PD studies. Appropriate application of the approach based on the knowledge of the test article and exposure history of the animals may reduce the number of nonhuman primates used to support preclinical studies of protein therapeutics.
Acknowledgments
The authors thank the following individuals for their excellent technical support: Abu Siddiqui and Monica Keen for conducting the ADA analyses, Paula Gegwich for mAb A concentration bioanalysis, Brian Jones for IL-17 concentration analysis, and Trina Jiao for PK analysis. We specifically acknowledge Ke Li and Sylvia Zhao for coordinating and managing the in-life portion of the PK study. We thank Kenneth D. Graham for critically reviewing and editing the manuscript.
Authorship Contributions:
Participated in research design: Han, Han Hsu, Davis.
Conducted experiments: Han, Gunn, Marini, Shankar.
Performed data analysis: Han, Gunn, Marini, Shankar.
Wrote or contributed to the writing of the manuscript: Han, Gunn, Marini, Shankar, Han Hsu, Davis.
Footnotes
- Received December 15, 2014.
- Accepted March 4, 2015.
Abbreviations
- ADA
- antidrug antibody to mAb A
- AUC
- area under the serum concentration–time curve
- AUCinf
- AUC from time 0 to infinity
- CL
- clearance
- ECL
- electrochemiluminescence
- Fc
- fragment, crystallizable, of antibody
- FcRn
- neonatal Fc receptor
- IL
- interleukin
- mAb
- monoclonal antibody
- PD
- pharmacodynamics
- PK
- pharmacokinetics
- t1/2
- terminal half-life
- Vz
- volume of distribution
- Copyright © 2015 by The American Society for Pharmacology and Experimental Therapeutics